soundcheck's Touch Toolbox 3.0

[darrell]

Fellow listeners, in my humble experience, do yourselves a very large favour and just bypass anything posted by this "member" - the propellor is clearly beginning to detach from the head - which will be delayed the longer attention is given / people reply to his inane posts.

Unless you are into seeing colours in higher definition.. or wish to pontificate on "improvements" that make zero difference to your listening experience..Or wish to watch the car crash happen in real time..

Total waste of breath and energy.

You're probably right. OK, for my part, factual corrections only in the future, no direct engagement. (But I do hope the propeller detachment/meltdown makes it on to Youtube sometime )

[JohnSwenson]

Thanks. I have tried and implemented many tweaks in my audio systems. But reducing buffer size just does not make sense to me at all....

My current thought is that buffer size affects the processor cache hit rate. The assumption here is that one of the main causes of noise in a computer (power rails, groundplane and radiated) is the main memory bus. These are very wide fast busses, when data goes over these busses the drivers produce huge current pulses which cause lots of noise. Each bit is not too bad but there can be a hundred bits switching at exactly the same time. So to me what makes sense is to cut down on main memory access.

Small buffers mean that its possible that the data can stay in the processor cache and never have to go out to main memory. There is a tradeoff here because a small buffer also means more interrupts and process switches which means more code gets swapped in and out of main memory.

This is something that is really hard to measure since OSes don't usually give you a processor cach hit rate graph you can look at. (of course the process of monitoring it would decrease the hit rate!)

John S.

[evdplancke]

It is often easy to quote some theory we knew from the past without verification. It also can be hard to verify the operating condition. For example, how do we know the error rate between the source and the DAC is zero? I am guessing that the error rate is probably not zero. If that is true, it is possible to do some tweaks on the source to reduce the error rate.

I think you are right. If I remember well the theory, given a signal to noise ratio and a type of modulation, we can calculate the probability of error (bit error rate). When we add noise to a signal, this probability is very close but not equal to 0. The question is: at what level the bit error rate becomes audible? can we hear a bit error rate of 1E-10 for instance?

Another question: is a DAC totally immune to analog noise that is added to its digital input (EMI, temperature noise,...)? I would tend to say no, if there is no galvanic isolation between input and output. Am I wrong?

If not, this could explain why source CPU load could have indirect incidence on analog output noise. Bit perfect means that the signal can be almost perfectly reconstructed when there is a good jitter rejection but the noise of the DAC circuitry should be added to the signal including potential noise transferred from the input if any.

[SBGK]

Soundcheck,

If you are going to put forward proposals for tweaks to the server side in TT4.0 then be careful not to post about it in this forum, my thread detailing such changes has been closed as the discussion was deemed stupid.

SBGK,

I'm one of the forum moderators over at the Squeezebox forums. I'm sorry
to say that you've caused so many complaints (including yourself
complaining) that I wonder if it's any good if you participate there.
Please re-consider your future postings twice: are they worth the hassle
they cause? All the discussions you have contributed to in a dauntless way
ended in personal attacks. I'd hate to have to ban you. But your recent
contributions were not helpful to the overall discussion culture in the
forums.

Thanks in advance!

I mean, I now have a superb music system, don't understand why other people don't want the same.

my reply

michael,

it's not just me who get's attacked, it seems to be the culture. Not invented here - then not welcome. I haven't seen this any where else.

Cheers

[Phil Leigh]

I think you are right. If I remember well the theory, given a signal to noise ratio and a type of modulation, we can calculate the probability of error (bit error rate). When we add noise to a signal, this probability is very close but not equal to 0. The question is: at what level the bit error rate becomes audible? can we hear a bit error rate of 1E-10 for instance?

Another question: is a DAC totally immune to analog noise that is added to its digital input (EMI, temperature noise,...)? I would tend to say no, if there is no galvanic isolation between input and output. Am I wrong?

If not, this could explain why source CPU load could have indirect incidence on analog output noise. Bit perfect means that the signal can be almost perfectly reconstructed when there is a good jitter rejection but the noise of the DAC circuitry should be added to the signal including potential noise transferred from the input if any.

Which is why it is important to galvanically isolate the Touch. There are three routes in:
1) the spdif coax connection - noise can be transferred back from the DAC - use Toslink or at least use a DAC that has a transformer isolated input stage

2) the ethernet connection - use wireless, or use a properly isolated ethernet cable (not screened at both ends) or use an ethernet isolater like this: http://www.blackbox.co.uk/product/sp426a/4/6/1/
3) the power supply

[evdplancke]

Which is why it is important to galvanically isolate the Touch. There are three routes in:
1) the spdif coax connection - noise can be transferred back from the DAC - use Toslink or at least use a DAC that has a transformer isolated input stage

2) the ethernet connection - use wireless, or use a properly isolated ethernet cable (not screened at both ends) or use an ethernet isolater like this: http://www.blackbox.co.uk/product/sp426a/4/6/1/
3) the power supply

Thank you for the advice. If I understand you well, if those 3 conditions are not perfectly met,you admit implicitly that some kind of noise can be transmitted from the source to the DAC output.

However these consideration seems to be based on the assumption that the noise is coming from external sources (ground noise from cables and power supply) and not from the signal path. But considering that a digital signal is a perfectly square wave + noise, we should also consider that some of the noise from the signal path can also be transferred to DAC output. Sources of this "added" noise are typically jitter, temperature noise and any kind of signal interference caused by PSU, CPU processing,...
Even with signal reshaping, the return path for this noise should be isolated from DAC output. In the same logic, I don't know how optocouplers work, but do they produce perfect square waves at their electrical output as well? if not, they would also transfer the noise from the signal path, even if they provide perfect galvanic isolation.

I don't want to raise polemic, but from my point of view, there are so many unanswered questions about handling of the noise in the reconstruction of the analog signal that I can not admit the "bit perfect" argument alone to conclude that a digital source noise has no impact on analog output noise.

And if I remember well Phil, you are the only one on this forum that objectivated by quantitative measurements that tt3.0 mods have an incidence on DAC output signal (weither it is audible or not is another topic).

[soundcheck]

Soundcheck,

If you are going to put forward proposals for tweaks to the server side in TT4.0 then be careful not to post about it in this forum, my thread detailing such changes has been closed as the discussion was deemed stupid.

Look. There is slight difference between you and me. Sometimes I have the feeling you don't see that.

The whole Toolbox thing has been my very own idea and work.

Server and network tweaks do have quite a long history on my blog.

"WLAN off" as the first network tweak was one of my very first tweaks.

Then you'll find several others mods that also relate to the network and server.

And... ...surprise, surprise my threads still exist. And I'm not banned.

This TT3.0 thread only has meanwhile close to 100000 hits (probably one of the highest hitcounts/time in this forum).

Threads like that keep a forum going.

I do think TT4.0 wouldn't cause less attention.

It's not me who gotta problem.



Please don't forget:

All this discussion is not about your "constructive" and "partly" valuable contributions.


The message you received is crystal clear, isn't it:

It's all about attitude and respect. (Yep. Something like that exists, even in the virtual world.)

Look. You even started trying to accuse me of certain things recently. That's not what I call a nice attiude.
In my own forum I would have banned you for that kind of attitude.

I can tell you. The way you act won't let you make many friends.


If your own thread is not even able to survive the "audiophile" section over here @ SB forums, you better change your attitude or you look for a different place to post your findings.

BTW: I never filed a complaint. Interesting to see that others did.


I really hope that we can stop all these nasty discussions and go on with
more constructive part of the project.

Thx.

Enjoy.

[Munroe]

I did some listening tests of my own to see what quality level my SBT is at out of the box via analog outs. I have to be honest I didn't find a large difference between the same music on phono vs the SBT. There is enough that I will try a DAC possibly, as there was two suttle differences in the music I played.

As a background, the two sources consisted of a turntable -> phono preamp -> amp, then a SBT to amp. Both of these connected to the same length and brand of analog cable to keep things consistent. The LP was a newly released Let it Bleed by the Rolling Stones, 180 gram DSD remastered and an HDtracks high res FLAC in 24 bit/96 kHz. The SBT was set to the same volume level as the phono, and both albums were played simultaneously to allow switching back and forth.

The resulting audio while very similar, had two distinct differences. the analog turntable clearly being the winner, the only shortcoming I found with the SBT was a less natural sounding high frequency (lp sound was a bit more "sweet" if that's a phrase in audio.) and a lack of presence ( or air) to the music, though peculiarly a touch more bottom end. To make a long story short, the tambourine on You Can't always get what you want sounded like it was actually in front of me via the turntable, though while very clear, the same tambourine on the SBT sounded more like a tambourine recording.

Will an external DAC bring this unit closer to an analog sound I witness through a turntable? Perhaps; but I am going to do a second test using the TT 3.0 through the analog, to test the theory that these tweaks actually bring your music closer to an analog sound than the stock. Which I assume is the point of this thread. The tweaks are free unlike a DAC, so I will try these first.

[Phil Leigh]

I did some listening tests of my own to see what quality level my SBT is at out of the box via analog outs. I have to be honest I didn't find a large difference between the same music on phono vs the SBT. There is enough that I will try a DAC possibly, as there was two suttle differences in the music I played.

As a background, the two sources consisted of a turntable -> phono preamp -> amp, then a SBT to amp. Both of these connected to the same length and brand of analog cable to keep things consistent. The LP was a newly released Let it Bleed by the Rolling Stones, 180 gram DSD remastered and an HDtracks high res FLAC in 24 bit/96 kHz. The SBT was set to the same volume level as the phono, and both albums were played simultaneously to allow switching back and forth.

The resulting audio while very similar, had two distinct differences. the analog turntable clearly being the winner, the only shortcoming I found with the SBT was a less natural sounding high frequency (lp sound was a bit more "sweet" if that's a phrase in audio.) and a lack of presence ( or air) to the music, though peculiarly a touch more bottom end. To make a long story short, the tambourine on You Can't always get what you want sounded like it was actually in front of me via the turntable, though while very clear, the same tambourine on the SBT sounded more like a tambourine recording.

Will an external DAC bring this unit closer to an analog sound I witness through a turntable? Perhaps; but I am going to do a second test using the TT 3.0 through the analog, to test the theory that these tweaks actually bring your music closer to an analog sound than the stock. Which I assume is the point of this thread. The tweaks are free unlike a DAC, so I will try these first.

Interesting as this is, you are comparing two wildly different masterings... so unfortunately it doesn't tell us very much other than you clearly prefer one mastering over the other (which is fine). It certainly says nothing about the relative sonic merits of the Touch and your turntable/catridge/phono pre-amp...

A much better test would be to compare a good rip of YOUR LP against the LP.
Whenever I have done this - approx 150 times - I have never heard the Touch sound worse - it has always sounded slightly better and the reason is because the vinyl playback suffers from the effects of acoustic feedback which the rip doesn't (unless it was done badly - i.e. with the sound playing through the speakers!)

[adamdea]

I think you are right. If I remember well the theory, given a signal to noise ratio and a type of modulation, we can calculate the probability of error (bit error rate). When we add noise to a signal, this probability is very close but not equal to 0. The question is: at what level the bit error rate becomes audible? can we hear a bit error rate of 1E-10 for instance?

There have been tests done to show bit perfect transmission to the dac.
Weiss dacs can do this. I belive that it has been confirmed on other dacs and in other ways that transmission is bit perfect. SPDIF contatins AFAIK error detection bits (buit not error correction).
I am not convinced that there is any reason why there shopul dnot be perfect recovery of all data provided that the trasnitions are not mixed up (which requires serval ns of jitter).
This old paper concludes that amplitude errors through noise are unlikely
http://www.scalatech.co.uk/papers/aes93.pdf . It refers to other papers on errors in transmission.

But let's suppose that there was an error of one in 10,000,000,000? if there are 705k music information bits per second for one channel of 16/44, that suggest that one sample every 4 hours might be somewhat wrong (depending on whether it was the least or most significant bit). I can't see that this would account for any audible effect because any effect on the output would only last an instant and would probably be filtered out by the Anti imaging filter- if one sample value was significantly changed it would presumably have the effect once decoded of creating a very very high frequency sound which would eb filtered out.

I had a look for an estimate of the bit error rate on spdif. I noted one forum where someone asserted the figure 10^-12 or 100 times lower than your figure.